Rewriting Excel, PowerPoint, and Word for Agents
Comments on “Rewriting Excel, PowerPoint, and Word for Agents”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
Compound has rewritten Excel, PowerPoint and Word as headless engines in Rust, so AI agents can calculate, edit, print and visually check office files on servers without Microsoft Office.
Why it ranked
No. 23 of 62 with a score of 0.64. The panel rated it highest on new information (0.92) and substantiated (0.83), lowest on audience breadth (0.61).
Every answer behind this is in the table under Score.
Score
0.64total under rubric v2.2
- Significance0.63
- Reader interest0.80
- Audience breadth0.61
- Practical relevance0.72
- Substantiated0.83
- Technical depth0.82
- New information0.92
- Counts against: Promotional0.22
- Counts against: Overclaim0.38
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| Significance | 0.72 | 0.45 | 0.72 | 0.63 | 0.27 | +0.24 | +0.151 |
| Reader interest | 0.78 | 0.72 | 0.91 | 0.80 | 0.19 | +0.26 | +0.209 |
| Audience breadth | 0.55 | 0.60 | 0.67 | 0.61 | 0.12 | +0.13 | +0.079 |
| Practical relevance | 0.71 | 0.60 | 0.86 | 0.72 | 0.26 | +0.22 | +0.159 |
| Substantiated | 0.91 | 0.72 | 0.86 | 0.83 | 0.19 | +0.05 | +0.041 |
| Technical depth | 0.83 | 0.78 | 0.84 | 0.82 | 0.06 | +0.05 | +0.041 |
| New information | 0.93 | 0.85 | 0.99 | 0.92 | 0.14 | +0.05 | +0.046 |
| Promotional | 0.10 | 0.45 | 0.12 | 0.22 | 0.35 | −0.19 | −0.042 |
| Overclaim | 0.47 | 0.40 | 0.28 | 0.38 | 0.19 | −0.11 | −0.042 |
| Sum of contributions | +0.642 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.642 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-10-26/23/"><img src="https://rubric.hrih.si/badge/2026-10-26/23.svg" alt="Ranked #23 on Rubric, 26 Oct 2026" width="245" height="28"></a>Claim check
Claim check: 7 no source given · 1 couldn’t check. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
No source givenNumber
Compound's spreadsheet and document engines were built in about eight months using about a trillion LLM tokens.
The article gives no link or citation for it.
The article says “We built both engines in about eight months, using about a trillion LLM tokens”
The source says No source to compare with.
No source givenNumber
On a workbook calling each Excel function, Compound matches Excel's stored result on 873 of 874 formulas.
The article gives no link or citation for it.
The article says “Compound matches Excel’s stored result on 873 of 874 formulas”
The source says No source to compare with.
No source givenNumber
On 22,523 public Sheetpedia workbooks (148.9 million formula cells), 98.5% of Compound's results are bit-for-bit identical to Excel's.
The article gives no link or citation for it.
The article says “98.5% of Compound’s results are bit-for-bit identical to Excel’s”
The source says No source to compare with.
No source givenBenchmark result
On a 63-formula workbook of newer functions, Compound matches Excel on all 63 formulas while LibreOffice 26.8.1 matches on 39.
The article gives no link or citation for it.
The article says “Compound matches Excel on all 63 formulas. LibreOffice matches on 39”
The source says No source to compare with.
No source givenNumber
Across 66 workbooks, Compound opens files 11 to 35 times faster than Excel and recalculates a whole workbook 1.2 to 2.5 times faster.
The article gives no link or citation for it.
The article says “Compound opens files 11 to 35 times faster than Excel (geometric mean, depending on the set), recalculates a whole workbook 1.2 to 2.5 times faster”
The source says No source to compare with.
No source givenBenchmark result
Across the corpus, Compound's mean SSIM against Office is 0.940 for PowerPoint pages and 0.948 for Word pages, versus 0.886 and 0.848 for LibreOffice 26.8.
The article gives no link or citation for it.
The article says “Compound 909 0.940 86% 799 0.948 89%”
The source says No source to compare with.
No source givenNumber
Compound converts a typical deck to PDF in 25 to 110 milliseconds, while headless LibreOffice took 2.8 seconds for a deck and 1.2 seconds for a Word document.
The article gives no link or citation for it.
The article says “Compound converts a typical deck to PDF in 25 to 110 milliseconds and a Word document in under 100”
The source says No source to compare with.
Couldn’t checkGeneralization
Claude's PowerPoint skill builds decks with the PptxGenJS library and renders them with LibreOffice to check the result.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “its PowerPoint skill builds decks with the PptxGenJS library and renders them with LibreOffice to look at the result”
The source says No passage could be compared.
Source cited by the article: github.com/anthropics/skills/blob/main/skills/pptx/SKILL.md
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 3 pts, not on front page
Recorded for context only, checked 26 Oct 2026, 07:27 UTC. It played no part in the ranking.